LLMs · Baidu
Unlimited OCR (3B-A0.6B)
- Parameters
- 3.3B (0.6B active)
- Architecture
- Mixture of experts
- Max context
- 32K tokens
- License
- mit
- Released
- 2026-06
How to run it
Original weights · Hugging Face
Full-precision files from Baidu, for fine-tuning, vLLM, Transformers.
Memory needed
Weights plus KV cache and runtime overhead. Leave a little extra free for the operating system and other apps.
| Quantization | Quality | Weights | Total at 8K context | Total at 32K context | Download |
|---|---|---|---|---|---|
Q2_K |
Low | 1.3 GB | 2.4 GB | 3.9 GB | All files (opens in a new tab) |
Q3_K_M |
Low | 1.6 GB | 2.7 GB | 4.2 GB | All files (opens in a new tab) |
Q4_K_M |
Good | 2.0 GB | 3.1 GB | 4.7 GB | All files (opens in a new tab) |
Q5_K_M |
High | 2.4 GB | 3.5 GB | 5.0 GB | All files (opens in a new tab) |
Q6_K |
High | 2.7 GB | 3.8 GB | 5.4 GB | All files (opens in a new tab) |
Q8_0 |
Very high | 3.5 GB | 4.6 GB | 6.2 GB | All files (opens in a new tab) |
F16 |
Full | 6.7 GB | 7.8 GB | 9.3 GB | Original weights (opens in a new tab) |
Best hardware for Unlimited OCR (3B-A0.6B)
The least expensive GPU in our catalog that runs it without offloading is the Intel Arc A380 (launch price $139) at Q8_0.
| Grade · points | Hardware | Quant | Memory | Speed | Fit |
|---|---|---|---|---|---|
| 73 ptsGreat (A) · 73 of 100 points: Model quality 18.0/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15 | Intel Arc B580 | F16 | 8.4 GB / 12.0 GB | 83.6 tok/s | Runs well |
| 73 ptsGreat (A) · 73 of 100 points: Model quality 18.0/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15 | NVIDIA GeForce RTX 3060 12GB | F16 | 8.4 GB / 12.0 GB | 86.4 tok/s | Runs well |
| 73 ptsGreat (A) · 73 of 100 points: Model quality 18.0/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15 | NVIDIA GeForce RTX 4070 SUPER | F16 | 8.4 GB / 12.0 GB | 121 tok/s | Runs well |
| 72 ptsGreat (A) · 72 of 100 points: Model quality 18.0/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 14.5/15 | NVIDIA GeForce RTX 4060 | Q8_0 | 5.0 GB / 8.0 GB | 142 tok/s | Runs well |
| 71 ptsGood (B) · 71 of 100 points: Model quality 18.0/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 13.0/15 | Apple M4 Pro (24 GB) | F16 | 9.1 GB / 16.1 GB | 56.4 tok/s | Runs well |
| 70 ptsGood (B) · 70 of 100 points: Model quality 18.0/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 12.4/15 | AMD Radeon RX 9070 XT | F16 | 8.6 GB / 16.0 GB | 132 tok/s | Runs well |
| 70 ptsGood (B) · 70 of 100 points: Model quality 18.0/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 12.4/15 | NVIDIA GeForce RTX 4060 Ti 16GB | F16 | 8.6 GB / 16.0 GB | 69.1 tok/s | Runs well |
| 70 ptsGood (B) · 70 of 100 points: Model quality 18.0/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 12.4/15 | NVIDIA GeForce RTX 4070 Ti SUPER | F16 | 8.6 GB / 16.0 GB | 161 tok/s | Runs well |
| 70 ptsGood (B) · 70 of 100 points: Model quality 18.0/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 12.4/15 | NVIDIA GeForce RTX 4080 SUPER | F16 | 8.6 GB / 16.0 GB | 177 tok/s | Runs well |
| 70 ptsGood (B) · 70 of 100 points: Model quality 18.0/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 12.4/15 | NVIDIA GeForce RTX 5060 Ti 16GB | F16 | 8.6 GB / 16.0 GB | 108 tok/s | Runs well |
Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.